Reflection Unveils Beam, a Western Answer to China's Open Models
Reflection AI has officially unveiled Beam, its first frontier, open-weight AI model. The two-year-old startup claims Beam matches the performance of leading Chinese open models on advanced reasoning benchmarks at dramatically lower costs β a claim that could intensify the race to build a Western counterweight to DeepSeek, Qwen, and Z.ai.
The announcement confirms reporting from Axios over the weekend that the startup was nearing a launch. Reflection shared additional details in a lengthy blog post on Monday, describing Beam as a text-only mixture-of-experts model trained on high-compute reinforcement learning to excel at reasoning, coding, and agentic tasks at "a fraction of the token cost and inference time compute" of rivals.
Model Architecture and Specifications
Beam is a 501-billion-parameter model with 23 billion active parameters. It was pre-trained on 23.8 trillion tokens and features a 1-million-token context window. For comparison, Z.ai's GLM-5.2 has roughly 744 billion total parameters with 40 billion active.
Reflection's performance claims have not been independently verified. However, on advanced reasoning benchmarks, the company says Beam scores on par with Z.ai's GLM-5.2 and outperforms today's leading Western open models while using "3β4x less inference compute." Reflection bills it as a "workhorse model" for enterprises, the public sector, and developers.
A Crowded Competitive Landscape
Reflection is positioning itself against closed labs like Anthropic and OpenAI, against popular open models from Chinese developers, and against Western players such as Mistral, Meta, and Cohere. Its most direct U.S. rival may be Inkling, the open model from Mira Murati's Thinking Machines Lab released in July. Reflection's own benchmarks show Beam outscores Inkling on four coding tests where both report results β though Inkling is multimodal, while Beam is text-only.
Funding, Compute, and Backing
Reflection was founded in 2024 by two former Google DeepMind researchers and has raised roughly $4.7 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners, according to PitchBook. Its last round valued the company at a $25 billion pre-money valuation.
The startup has also been locking up compute β a key ingredient for training frontier models capable of luring customers away from Anthropic and OpenAI's closed models, as well as the cheaper open-weight models from Chinese labs. This summer, Reflection signed deals collectively worth more than $7 billion with SpaceX and Nebius to secure access to Nvidia's GB300 chips through 2029.
The "AI Factory" Pitch
Reflection is aiming Beam and future models at enterprises and sovereign nations. The pitch centers on building "AI factories" β a product that would let institutions construct their own customized, local AI systems by training Reflection's models on their proprietary data. Nvidia CEO Jensen Huang, whose company backs Reflection, has long championed the "AI factory" concept and pushed to strengthen the open AI ecosystem β a vision that would also benefit Nvidia, whose GPUs would power those systems.
Axios reported that hedge funds and trading firms are among those eager to build such systems. Reflection has already begun testing the sovereign AI factory concept through a partnership with Shinsegae Group in South Korea.
Release Plans
Reflection says it will release Beam's weights and full technical details this month, with distribution through hyperscalers and neoclouds and integrations across open-source libraries at launch.
via TechCrunch AI